ORIGINAL ARTICLE

Assessment of the impact of climate change on surfacehydrological processes using SWAT: a case study of Omo-Giberiver basin, Ethiopia

Shiferaw Eromo Chaemiso1 • Adane Abebe2 • Santosh Murlidhar Pingale2

Received: 29 October 2016 /Accepted: 2 November 2016 / Published online: 16 November 2016

� Springer International Publishing Switzerland 2016

Abstract In the present study, an ArcSWAT model was

utilized to simulate the hydrological responses due to land

use and climatic changes in the Omo-Gibe river basin,

Ethiopia. The performance of the model was evaluated

through sensitivity, uncertainty analysis, calibration and

validation. The most sensitive parameters were identified

which are governing on surface runoff generation processes

in the selected basin. The calibration and validation of the

model was done using SWAT-2005. Also, the sensitivity

and uncertainty analysis was performed using the SUFI-2

and SWAT-CUP algorithm. The model results revealed

that a good performance during the calibration

(R2 = 72.4%, NSE = 62.6% and D = 14.37%) and vali-

dation (R2 = 68.1%, NSE = 68% and D = 4.57%). SUFI-

2 algorithm gave good results in minimizing the differ-

ences between observed and simulated flow in the Great

Gibe sub-basin. The studies show that there is an overall

increasing trend in future annual temperature and signifi-

cant variation of monthly and seasonal precipitation from

the base period 1985–2005. Also, the annual potential

evapotranspiration shown increasing trend for future cli-

mate change scenarios. Similarly, the surface water

decreases in terms of mean monthly discharge in the dry

season and increases in the wet season. The percentage

change in future seasonal and annual hydrological vari-

ables was shown increasing trends. Therefore, this study

found that SWAT can be effectively used for assessing the

water balance components of a river basin in Omo-Gibe

basin, Ethiopia.

Keywords SWAT � ArcGIS � Climate change � Water

balance � Omo-Gibe basin � Ethiopia

Introduction

Precipitation is one of the most important hydrological

variables of the basin (Sevruk et al. 1998). It greatly

influences the amount of water flowing through the water

cycle and water availability. Similarly, temperature is

another important parameter in assessing the climate

change impact on water resources system. According to

climate model predictions, using several scenarios of

greenhouse gas emissions, global mean temperature prob-

ably will increase from 1.1 to 6.4 �C in the next 100 years

(IPCC 2001). This means there will be an increase of

extreme weather events as well as change in the precipi-

tation and atmospheric circulation patterns. Shift in pre-

cipitation and temperature patterns affects the hydrology

process and availability of water resources. Warmer tem-

perature increases the water-holding capacity of the air and

thus increase the potential evapotranspiration (PET),

reduce soil moisture and decrease groundwater reserves,

which ultimately affects the river flows and water avail-

ability (IPCC 2001b). Changes in precipitation and evap-

oration have a direct effect on the groundwater recharge.

More intense precipitation and longer drought periods,

which are considered to be expected impacts of climate

& Santosh Murlidhar Pingale

pingalesm@gmail.com; santosh.murlidhar@amu.edu.et

Shiferaw Eromo Chaemiso

shiferomo@gmail.com

1 Department of Hydraulics and Water Resources Engineering,

Arba Minch Institute of Technology, Arba Minch University,

P.O. Box 21, Arba Minch, Ethiopia

2 Department of Water Resources and Irrigation Engineering,

Arba Minch Institute of Technology, Arba Minch University,

P.O. Box 21, Arba Minch, Ethiopia

123

Model. Earth Syst. Environ. (2016) 2:205

DOI 10.1007/s40808-016-0257-9

changes for most of the land areas of the world, could

cause reduced groundwater recharge (IPCC 2001a).

Climate change affects the function and operation of

existing water infrastructures including hydropower,

structural, drainage and irrigation systems as well as water

management practices. The existing land and water

resource system of the area is adversely affected by the

rapid growth of population, deforestation, surface erosion,

and sediment transport and climate change impacts. Cli-

mate change increases the vulnerability of poor people,

affects their health and livelihoods and undermines growth

opportunities crucial for poverty reduction (AfDB 2003).

Extreme events due to anthropogenic climate change

would cause forced migration and human resettlement

resulting in the damage of the social cohesion including the

loss of human lives and physical properties. Therefore, it is

very important to quantify such impacts in order to identify

the variation in decisions and thereby minimize the

potential damage magnitude of climate change on a local

and regional scale (Chaulagain 2006). This information can

point out the sensitive area that might be affected by cli-

mate change and help in designing a proper plan to manage

water resources in changing climate scenarios (Dincer et al.

2009).

Climate change is one of the crosscutting issues whose

impacts need to be considered in any planning. It is con-

sidered as one of the risks that might face the future

development within the basin, which has to be considered.

All these aspects affect livelihoods in the basin but have

not received attention in the planning for the future water

allocation and design of water infrastructures yet (Kim

et al. 2008). It also indicate that developing countries like

Ethiopia will be more vulnerable to climate change due to

their economic, climatic and geographic settings (IPCC

2007). Also finds the population at risk of increased water

stress in Africa. Due to these, the agricultural production

from rainfed agriculture could be reduced in some coun-

tries that mainly depended on rainfed agriculture. Hence,

assessing the impact of climate change on hydrological

components like on streamflow, soil moisture, groundwater

and other hydrological component, essentially involves

projections of climatic variables (e.g. precipitation, tem-

perature, humidity etc.) at a global and local scale were

very crucial to investigate the level of climate change

impact on water resources (Ghosh and Mujumdar 2009).

Over the last decades, temperature in Ethiopia has

increased at about 0.37 and 0.28 �C per decades (NMA

2007; McSweeney et al. 2008). The increase in minimum

temperatures is more pronounced with roughly 0.4 �C per

decade. Precipitation, on the other hand, remained fairly

stable over the last 50 years when averaged over the

country. However, the spatial and temporal variability of

precipitation are high (NMA 2007). Temesgen et al. 2006

studied vulnerability of region to climate change, which

indicated that the net effect of sensitivity, exposure, and

adaptive capacity are different across the region. Afar,

Somali, Oromia, and Tigray are relatively more vulnerable

to climate change than the other regions. The vulnerability

of Afar and Somali is attributed to their low level of rural

service provision and infrastructure development. Tigray

and Oromia’s vulnerability to climate change can be

attributed to the regions higher frequencies of droughts and

floods, lower access to technology, fewer institutions, and

lack of infrastructure. SNNPR’s lower vulnerability is

associated with the region’s relatively greater access to

technology and markets, larger irrigation potential, and

higher literacy rate (Temesgen et al. 2006).

The majority of previous studies on the potential impact

of climate change on availability of water were limited (i.e.

Nile, Awash, Rift Valley river basins). Therefore, this

study tried to investigate how changes in terms of tem-

perature and precipitation might translate into changes in

streamflow, surface water availability for irrigation and

other hydrological parameters using outputs from a regio-

nal climate models (RCM). In A1B scenario of RCM

output is used to evaluate hydrological impacts on Omo-

Gibe basin water balance. Also, Omo-Gibe river basin area

is pastoral and agro-pastoral, the area is highly vulnerable

to climate change and adaptation strategy is vital. Hence,

new strategies for effective use of the water in the basin

particularly in the Omo River basin is needed for water

development and management to prevent water scarcities.

Therefore, the main objective of the study is to assess the

impacts of climate change on hydrological processes such

as water balance components of the basin using SWAT

model in Omo-Gibe river basin, Ethiopia.

Materials and methods

Study area

The Omo-Gibe river basin has an area of 79,000 km2,

covering parts of two national regional states, the Southern

Nations and Nationality Peoples Region (SNNPR) and

Oromia (Fig. 1). It lies between 4�300 to 9�000N Latitude

and 35�000 to 38�E longitude. The total mean annual flow

from the river basin is estimated at about 16.6 billion cubic

meter (BMC). Gibe River is known as the Omo River in its

lower reaches, south-westwards from the confluence with

the Gojeb River. The main tributaries of Omo-Gibe River

Basin are from north-east Walga, Wabe, Wariness and

Derghe and south-west Tunjo (called Gilgel Gibe as sta-

tion) and Gilgel Gibe Rivers and the west Gojeb River.

From the eastern side of the middle and lower Omo-Gibe

catchment the Sana, Soke, Wabe, Deme and Zage rivers

205 Page 2 of 15 Model. Earth Syst. Environ. (2016) 2:205

123

are the main tributaries. It is an enclosed river basin that

flows into the lake Turkana in Kenya, which forms its

southern boundary. The western watershed is the range of

hills and mountains that separate the Omo-Gibe basin from

the Baro-Akobo basin. To the north and north-west the

basin is bounded by the Blue Nile basin with small area in

the northeast bordering the Awash basin.

Basically, climate of Ethiopia is classified into five cli-

mate zones based on the altitude and temperature. Namely,

Wurch (cold climate and the altitude is more than 3000 m),

Dega (temperate like climate of high land and the altitude

is between 2500–3000 m), Woina-Dega (warm climate the

altitude is between 1500–2500 m), Kola (hot and arid type

of climate and the altitude is less than 1500 m) and Bereha

(hot and hyper arid type of climate) (NMSA 2001). The

elevation of the study area is lying between 719 to 3086 m

above the mean sea level (masl). Annual rainfall varies

from 400 mm in the extreme south low land to 1900 mm

in the high land with the average being 1140 mm. The

mean annual temperature in the basin varies from less than

17 �C in the west highlands to cover 29 �C in the south low

lands (MoWR 1996). The high lands areas have elevations

of 3625 masl on mount Ghuge while the low land areas fall

in the altitudes of up to 235 masl. Steep slopes with dis-

sected hills characterize the highlands while the low lands

are characterized by relatively gentle and undulating

slopes. The soils in the basin is highland soils and low land

soils. High land soils are the vast and majority of the parent

materials in the high lands are igneous rocks which under

the influence of a pluvial regime and relatively warm

temperatures have weathered to form deep well drained

clay soils with moderate natural fertility over almost all the

high land areas. Low land soils: soil derived primarily from

siliceous basement rocks or alluvial soils reworked from

soils derived from these rocks are generally coarse grained

and have a low nutrients status. Other alluvial soils can

have moderate to high fertility. However in the low lands,

cultivation is generally excluded by climate (MoWR 1996).

Data collection and database preparation

The meteorological (temperature, precipitation, relative

humidity, sunshine hour and wind speed), hydrological,

land use, and soil data was collected from Ministry of

Water Resource and Energy office (MoWE) and National

Meteorological Agency (NMA). Daily rainfall data of

20 years were collected from more than 20 gauging sta-

tions in the Omo-Gibe River basin (Fig. 2). The 19 years

(1986–2005) of daily discharge data was used for this

study. The digital elevation model (DEM) was obtained

Fig. 1 The index map of the Omo-Gibe river basin, Ethiopia

Model. Earth Syst. Environ. (2016) 2:205 Page 3 of 15 205

123

from the NASA Shuttle Radar Topographic Mission

(SRTM) with a resolution of 90 m 9 90 m (Fig. 3a). Slope

was generated from DEM using ArcGIS 9.3 (Fig. 3b). It is

used to analyze the drainage pattern of the watershed,

slopes, stream lengths, and widths of channel within the

watershed. The soil and land use data was obtained from

MoWE using GIS (Fig. 4). The areal coverage of the

LULC and slope of the basin are presented in Tables 1 and

2. The raw data was processed for consistencies and

homogeneities, filling of missed data, and quality check.

The inconsistency of a record was checked by double mass

curve technique (Subramanya 1998).

The RCM data has been obtained from the International

Water Management Institute (IWMI) for the different time

periods (2030s and 2090s) of A1B scenario. The bias

correction was applied to the RCM data of future precipi-

tation and temperature. The climate data (GCM) are

downscaled into fine spatial resolution RCM data of A1B

scenario. The nonlinear correction for each daily precipi-

tation P and a linear bias correction for temperature T was

applied and transformed to a corrected p* and T*,

respectively, using:

p* ¼ aPb and T* ¼ a � T þ b: ð1Þ

The coefficient a and b are determined iteratively.

Configuration of hydrological model using SWAT

SWAT (Soil and Water Analysis Tool) model was devel-

oped by USDA in 1995. It is semi-distributed physically

based hydrological model operating on a daily time step.

SWAT has been successfully applied numerous times for

long-term continuous simulations of flow, soil erosion, and

sediment and nutrient transport in watersheds of different

sizes, and having different hydrologic, geologic, and cli-

matic conditions. The model has been useful to study

impacts of certain climate changes on long term water

yields, and the impacts of certain management scenarios on

long term sediment and nutrient loads. The SWAT model

embedded in ArcGIS interface and works on the principle

Fig. 2 Rainfall gauging stations located in Omo-Gibe river basin

Fig. 3 a DEM and b slope (%) of the study area

205 Page 4 of 15 Model. Earth Syst. Environ. (2016) 2:205

123

of dividing the watershed into a number of homogenous

sub-basins that is called hydrologic response units (HRUs)

which have unique soil, slope and land use properties

(Neitsch et al. 2004; Arnold and Fohrer 2005). Hydrolog-

ical cycle as simulated by SWAT is based on the water

balance equation

Swt ¼ SwoXt

i¼1ðRday � Qsurf � Ea � Sseep � QgwÞ; ð2Þ

where, Swt is the soil water content (mm), Swo is the initial

soil water content (mm), Qsurf is the surface runoff (mm),

Rday is the amount of precipitation (mm), Ea is the Amount

of evapotranspiration (mm), Wseep is the soil infiltration i

(mm), and Qgw is the return flow (mm).

In this study, the SCS curve number method was used to

estimate surface runoff. FAO Penman–Monteith method

was adopted to calculate the daily potential evaporation

(Allen et al. 1998).

Calibration and validation

The stations from a period of 1985–1998 has been used for

model calibration and 1999–2005 used for model valida-

tion. After calibrating the models at the main gauging

station, comparison and selection of the best model were

made. The model performance was evaluated at both

catchments, then simulation of the flow was made on

Gilgel Gibe three dam sites for optimizing the impact of

climate change on extreme seasonal flows. The future cli-

mate change impact was assessed on the availability of

water balance in the catchments. The detailed methodology

adopted in the present study has been presented in the

Fig. 5.

Fig. 4 a Delineated LULC and b soil map of the study area

Table 1 The statistics of land use land cover

No. Land use type SWAT code Area (ha) % Coverage

1 Wood land AGRR 3,179,060 44.83

2 Cultivation AGRC 300,224 4.23

3 Natural forest PAST 3,148,696 44.4

4 Shrubs land WATR 840 0.81

5 Grass land SPAS 320,224 4.52

6 Afro-alpine FRSE 31,412 0.44

7 Plantation COFF 4920 0.65

8 Water URBN 1572 0.12

Table 2 Slope class of Omo-Gibe Basin

No. Classes slope Range in % Area (ha) % of coverage

1 1 0–5 3,082,284 43.46

2 2 5–15 2,281,008 32.16

3 3 15–999 1,654,228 23.33

Model. Earth Syst. Environ. (2016) 2:205 Page 5 of 15 205

123

In this study, the SWAT-CUP was used to perform

uncertainty and sensitivity analysis during calibration and

validation of the SWAT. The SUFI-2 algorithm was used

in SWAT calibration and validation which represents the

uncertainties of all sources (i.e. input, model and output)

(Yang et al. 2008). It can perform parameter sensitivity

analysis to identify those parameters that contributed most

to the output variance due to input. A comprehensive

description on the SUFI-2 can be found in (Abbaspour

et al. 1997).

The model performance in simulating observed dis-

charge was evaluated during calibration and validation

by the three performance indicators: Nash and Sutcliffe

efficiency (NSE) (Nash and Sutcliffe 1970), coefficient

of determination (R2) and Percent difference/Relative

Volume Error (D). Their formulations are given as

follows:

NSE ¼ 1�Pn

i¼1 Qo � Qs½ �2Pn

i¼1 Qo � Qo

� �2 ð3Þ

R2 ¼Pn

i¼1 Qs � Qs

� �Qo � Q0

� �� �2Pn

i¼1 Qs � Qs

� �� �2 Pni¼1 Qo � Qo

� �� �2 ð4Þ

D ¼ 100%�Pn

i¼1 Qo �Pn

i¼1 QsPni¼1 Qo

� �; ð5Þ

where, Qo is the observed flow, Qs is the simulated flow,

QO is the average of observed flow and Qs is the average of

simulated flow.

Results and discussion

Sensitivity analysis

According to Lenhart et al. (2002), the sensitivity analysis

has been carried out for 27 parameters and the most sen-

sitive parameters were identified (Table 3). The results of

the sensitivity analysis showed that CN is the top rank for

flow yield, followed by soil evaporation compensation

factor, base flow alpha factor and so on (Table 3).

Uncertainty analysis using SWAT_CUP

There are various sources of uncertainties which is related to

data, model assumptions and GCM output. After finding the

sensitive parameters on streamflow simulation (Table 4), the

SUFI-2 algorithm was used in SWAT-CUP to calculate the

calibration and validation parameters. Then, the calibration

periods were defined from 1992 to 1996 and the validation

period from 1998 to 2000. SUFI-2 algorithms gave good

results in minimizing the differences between observed and

simulated flow in the Great Gibe sub-basin. The p-factor and

r-factor was found more than 60% (Table 5; Figs. 6, 7, 8, 9).

According to Bosch et al. (2004) found that SWAT stream

flow estimates for small watershed were more accurate using

high resolution topographic, land use, and soil data versus

low resolution data obtained from basins. Cotter et al. (2003)

report that DEM resolution was the most critical input for a

SWAT simulation of DEM, land use, and soil resolution to

obtain accurate flow, sediment, nitrate.

GCM output

Climate Model

Downscaled RCM

Model calibration/ validation

Water Balance

scenario Swat –cup calibration/ validation

Bias correction

Assessment of climate change impact on hydrology

Verify result

Meteor/c

al data Hydro/cal

data

DEM and

GIS data

Hydrological model (SWAT)

Sensitivity analysis

Fig. 5 Flow chart for the

general frame work of the study

205 Page 6 of 15 Model. Earth Syst. Environ. (2016) 2:205

123

Calibration and validation at Great Gibe, Abelti

and Gojeb sub-watershed

Calibration and validation processes were performed

using the Parasol and manual methods using SWAT2005,

and the SUFI-2 methods for SWAT-CUP. The calibration

was done using historical records of 10 years

(1985–1995) of Great Gibe gauging station. The curve

number (CN), Soil evaporation compensation factor, Base

flow alpha factor (ALPHA_BF), Threshold depth of water

in the shallow aquifer for re-evaporation to occur

(REVAPMN), Effective and hydraulic conductivity in the

0.00

200.00

400.00

600.00

800.00

1000.00

1200.00

Jan-

88

Jul-8

8

Jan-

89

Jul-8

9

Jan-

90

Jul-9

0

Jan-

91

Jul-9

1

Jan-

92

Jul-9

2

Jan-

93

Jul-9

3

Jan-

94

Jul-9

4

Jan-

95

Jul-9

5

flow

in m

3/s

Simulated MeasuredFig. 6 Calibration result of

average monthly simulated and

measured flow at the outlet of

the sub basin 5, where Great

Gibe gauging station is located

Table 3 Sensitivity ranking of parameters towards water flow

Rank Parameter symbol Description Mean relative Sensitivity

1 Cn2 Initial SCS runoff curve number 2.89E-01 High

2 Esco Soil evaporation compensation factor 2.51E-01 High

3 Alpha_Bf Base flow alpha factor 2.26E-01 High

4 Revapmn Threshold depth of water in the shallow aquifer for re-evaporation to occur 1.60E-01 Medium

5 Ch_K2 Effective hydraulic conductivity in the main channel alluvium 1.35E-01 Medium

6 Gwqmn Threshold depth of water in the shallow aquifer required for return flow to occur 1.01E-01 Medium

7 Sol_Z Depth from soil surface to bottom of layer 9.62E-02 Small

8 Sol_Awc Available soil water capacity (m/m) 9.33E-02 Small

Table 4 Maximum and

minimum boundaries of

the parameters and fitted values

after calibration

No. Parameter name SUFI-2 Lower bound Upper bound Final fitted value

Daily Monthly

1 r_CN2.mgt -25% 25% 0.85 0.75

2 r_SOL_AWC.sol -25% 25% 0.02 0.021

3 r_CH_K.rte 0 150 11.8 6.20

4 v_ALPHA_BF.gw 0 1 0.068 0.044

5 v_GWQMN.gw 0 5000 37.84 35.65

6 v_Esco.hru 0 1 0.01 0.01

7 V_REVAPMN.gw 0 500 0.04 0.012

r_ means relative change (%) and v_ means value change

Table 5 SUFI-2 algorithms

iterationIndex Calibration Validation

Nash-Sutcliff efficiency R2 Nash-Sutcliff efficiency R2

Monthly 0.64 0.69 0.71 0.72

Daily 0.642 0.66 0.6 0.68

Model. Earth Syst. Environ. (2016) 2:205 Page 7 of 15 205

123

main channel alluvium, Threshold depth of water in the

shallow aquifer required for return flow to occur

(GWQMIN) parameters were used to carry out the water

balance part of the calibration. The first eight parameters

were considered according to their relative sensitivity for

temporal flow calibration (Table 6). Each time the tem-

poral calibration is finalized, the water balance was also

checked since the adjustment of the base flow parameters

can also affect the surface runoff volume. The perfor-

mance of the calibrated model was carried out using

statistical indicators [i.e. Nash–Sutcliffe coefficient

(NSE), Coefficient of Determination (R2)]. In the vali-

dation, data for a period of three years from the same

station of the years 1996–2000 was used.

0100200300400500600700800900

Jan-92

Jan-93

Jan-94

Jan-95

Jan-96

Jan-97

flow

0100200300400500600700

Jan-98

Apr-9

8Jul

-98Oc

t-98Jan

-99Ap

r-99

Jul-99

Oct-99Jan

-00Ap

r-00

Jul-00

Oct-00

flow

observed

(a)

(b)

Best_Sim

observed Best_SimFig. 7 Monthly calibration

(a) and validation (b) resultsshowing the 95% prediction

uncertainty intervals along with

the measured discharge at Great

Gibe near to Abelti Gauging

station

(a)

(b)

0200400600800

1000120014001600

1/1/19

98

3/1/19

98

5/1/19

98

7/1/19

98

9/1/19

98

11/1/199

8

1/1/19

99

3/1/19

99

5/1/19

99

7/1/19

99

9/1/19

99

11/1/199

9

1/1/20

00

3/1/20

00

5/1/20

00

7/1/20

00

9/1/20

00

11/1/200

0

flow

observed Best_Sim

0200400600800

100012001400160018002000

1/1/19

92

4/1/19

92

7/1/19

92

10/1/199

2

1/1/19

93

4/1/19

93

7/1/19

93

10/1/199

3

1/1/19

94

4/1/19

94

7/1/19

94

10/1/199

4

1/1/19

95

4/1/19

95

7/1/19

95

10/1/199

5

1/1/19

96

4/1/19

96

7/1/19

96

10/1/199

6

flow

observed Best_Sim

Fig. 8 Daily calibration (a) andvalidation (b) results showingthe 95% prediction uncertainty

intervals along with the

measured discharge at Great

Gibe near to Abelti Gauging

station

205 Page 8 of 15 Model. Earth Syst. Environ. (2016) 2:205

123

As discussed in data analysis section, the flow was cal-

ibrated automatically by the model using the observed areal

precipitation, areal evapotranspiration and observed flow at

abelti gauging station. The calibration was performed for

8 years period from January 1, 1988 to December 31, 1995.

The performance of the model for both simulated and

measured flow has been presented in Table 7. The R2 was

found to be 0.724, NS = 0.626, and D = 14.34 which

shows the simulation’s very good correlation with the

gauged flow (Figs. 6, 7, 8). DEM resolution affects delin-

eation of the watershed and stream network. A decrease in

spatial resolution results in a decrease in the volume of

simulated stream flow. The DEM 90 m 9 90 m required to

achieve less than 14.34% error at calibration and 4.57% in

validation error of flow simulation in this basin. The vali-

dation was performed for 5 years period from January 1,

1996 to December 31, 2000. The results indicates that the

simulated flow has very good correlation with the gauged

flow (R2 = 0.681 and NSE = 0.68) (Table 7; Fig. 9). In

addition, regression plots also confirms measured and

simulated monthly streamflow in good agreement (Fig. 10).

Similar to the Great Gibe and Abelti sub watershed, the

model was calibrated and validated (January 1, 1996 to

December 31, 2000) for Gojeb sub watershed (Table 7).

The performance of the model was found very good for

simulated flows (R2 = 0.74 and NSE = 0.714). Also

regression plot confirms measured versus simulated

monthly streamflow (Fig. 11).

Water balance components on average annual basis

SWAT model can be effectively used for assessing the

water balance components of a river basin. Water balance

components before and after calibration such as surface

runoff, lateral flow, base flow and evapotranspiration show

in Fig. 12 and Table 8. The prediction of basin water

balance is based on the simulation result from 1985 to

2005. The estimated annual precipitation falling on the

basin is 1380.8 mm and the evaporation loss from the basin

is about 645 and 1407.5 mm, 1485.9, 1568 mm annual

precipitation and 659.9, 719.9, 770.9 mm the evaporation

loss for the 2030 s and 2090 s, respectively.

Assessment of climate change and water balance

in Gibe III

The results shows that increasing trend in future average

maximum and minimum temperature for most of the sub-

basins (Fig. 13a), but in the case of the precipitation, the future

condition exhibits a fluctuating trend i.e. in some of the sub-

0

100

200

300

400

500

600

700

800

Jan-

96

Apr

-96

Jul-9

6

Oct

-96

Jan-

97

Apr

-97

Jul-9

7

Oct

-97

Jan-

98

Apr

-98

Jul-9

8

Oct

-98

Jan-

99

Apr

-99

Jul-9

9

Oct

-99

Jan-

00

Apr

-00

Jul-0

0

Oct

-00

flow

in m

3/s

Simulated Measured Fig. 9 Correlation graph

between measured and validated

model results

Table 6 The percentage

change in annual hydrological

variables of future periods with

respect to the base period

Time GCM output date corrected A1B scenario Percentage of change

Hydrological component Baseline 2000s 2030s 2090s %D2030 %D2090

Precipitation (mm) 1407.5 1485.9 1568 5.6 11.4

Evapotranspiration (mm) 659.9 719.9 770.9 9.1 16.8

Surface runoff (mm) 66.86 71.72 83.6 7.3 25.0

Ground water flow (mm) 465.41 547.62 553.3 17.7 18.9

Lateral flow (mm) 113.9 114.35 121.09 0.4 6.3

Percolation below root zone (mm) 506.25 582.33 591.78 15.0 16.9

Transmission losses (mm) 7.92 8.46 8.84 6.8 11.6

Total water yield 696.83 725.24 749.15 4.1 7.5

Table 7 Performance of calibration and validation

Station Calibration Validation

NSE R2 D NSE R2 D

Abilate 0.626 0.724 14.34 0.676 0.681 4.57

Gojeb 0.644 0.705 -1.87 0.714 0.732 0.51

Model. Earth Syst. Environ. (2016) 2:205 Page 9 of 15 205

123

basin increasing and in other decreasing trend. This may

attributed to complex nature of precipitation processes and its

distribution in space and time. Figure 13d shows the trend

changeofprecipitation in2030–2040 for sub-basindecreasing

trend in Belg season and Bega and increasing Kiremt season,

but for long term 2091–2100 increasing in Bega and Kiremt

decreasing inBelg season.The average annual evaporation for

Gibe III shows an increasing trend from short term

(2030–2040) to long term (2091–2100) in A1B scenario

(Fig. 14a). The results of the analysis of the mean seasonal

temperature record clearly showed that all the stations had

positive trends in Bega, Belg and Kiremt seasons. The tem-

perature is showing an increasing trend in Gibe III Fig. 14b.

Impact of climate change on water availability was

assessed based on climate change scenarios for the water-

shed. The future projected precipitation and temperatures

were downscaled using A1B scenario for the two future

climate periods (2030s and 2090s). The SWAT simulation

for the 1985–2005 period was used as a baseline period

against which the climate impact was assessed. The daily

precipitation and minimum and maximum temperature of

the future two periods: 2030–2040 and 2090–2100 were

directly used as input for SWAT. Other climate variables

as wind speed, solar radiation, and relative humidity were

assumed to be constant throughout the future simulation

periods based on this scenario. In Fig. 15b and Table 9

y = 1.4958x - 53.486 R² = 0.7243

-200

0

200

400

600

800

1000

1200

0 200 400 600Si

mul

ated

stre

amflo

w(m

m)

Measured streamflows(mm)

(a) Calibra�on period

y = 1.067x - 3.9536 R² = 0.6814

0

100

200

300

400

500

600

700

800

0 200 400 600 800

Sim

ulat

ed st

ream

flow

s(m

m)

Measured streamflows(mm)

(b) Valida�on period Fig. 10 Regression plots of

measured versus simulated

monthly streamflow relative to

lines for the a Calibration period(1988–1995), and b validation

period (1996–2000)

y = 0.9677x - 28.038 R² = 0.7056

-100

0

100

200

300

400

500

600

0 200 400 600

Sim

ulat

ed st

ream

flow

s(m

m)

Measured streamflows(mm)

(a) Calibr�on period

y = 1.2134x - 37.297 R² = 0.7365

-100

0

100

200

300

400

500

600

700

0 100 200 300 400 500

Sim

ulat

ed S

trea

mflo

ws(

mm

) Measured streamflow (mm)

(b) valida�on period Fig. 11 Regression plots of

measured versus simulated

monthly streamflow relative to

1:1 lines for the a calibration

period (1988–1995), and

b validation period (1996–2000)

0200400600800

1000120014001600

RF (mm)

ET (mm)

Sur_Q

(mm)

Lat_Q

(mm)

GW_Q

(mm)

WYLD

(mm)

SW (m

m)

PERC (mm)

LOSS

Hydrological component

Wat

er B

alan

ce in

(mm

)

Before Calibration After CalibrationFig. 12 Water balance in the

basin

205 Page 10 of 15 Model. Earth Syst. Environ. (2016) 2:205

123

above the percentage change in seasonal and annual

hydrological variables of future periods was showing that

increasing trends.

Impact of climate change on future surface water

availability

Changes in climate (precipitation and temperature) can

have a significant effect on the quality and balance of

surface water. After calibration and validating the

hydrological model with the historical records and bias

correction of the meteorological variables of downscaled

GCM output, the next step is the simulation of river flows

in the watershed. Also, subsequently the hydrological

model was used to identify possible trend in the simulated

river flow. Impact of climate change on water availability

was assessed based on climate change scenarios using

RCM model output which was downscaled from GCM

model for the watershed. Based on this, the hydrological

impact of the Omo-Gibe basin was analyzed using the

SWAT model for the future A1B scenario (2030s and

2090s).

The SWAT simulation for the 1985–2005 period was

used as a baseline period. Even if, it is definite that in the

future land use changes will also take place. This was also

assumed to be constant in this study to investigate the

impact with respect to the change in the climate variables

keeping all other factors constant. Therefore, Table 9 and

Fig. 16 show that percentage change of precipitation and

evaporation for future long time mean annual change

increasing trend from -2.58 to 2.49% precipitation 2030s

to 2090s and 5.52–12.35% evaporation 2030s–2090s. The

Fig. 13 a Minimum and b maximum temperature, % change; c monthly percentage change in precipitation Gibe III and d seasonal variation of

precipitation for Gibe III

Table 8 Water balance components and model result before and after calibration

Period RF (mm) ET (mm) Sur_Q (mm) Lat_Q (mm) GW_Q (mm) WYLD (mm) SW (mm) PERC (mm) Loss

Before 1380.8 645.8 136.30 108.99 456.25 692.47 16.42 489.19 9.07

After 1270.2 609.7 132.82 93.72 404.43 622.76 22.10 433.81 8.21

ET evapotranspiration, SURQ surface runoff, LATQ lateral flow into stream, GW_Q groundwater contribution to stream flow, WYLD

SURQ ? LATQ ? GW_Q-LOSSES, SW soil water, PERC percolation below root zone (groundwater recharge)

Model. Earth Syst. Environ. (2016) 2:205 Page 11 of 15 205

123

future precipitation and evaporation shown both decrease

for January, April, and precipitation decrease in Mar and

Nov (Fig. 16). The other months showed the increase

trend on the whole. This is mainly because of the domi-

nant impact of the average seasonal precipitation. In

general, the mean seasonal and annual change of soil

moisture varies with precipitation and evapotranspiration.

Additionally, the future soil moisture will also vary with

changing water yields, surface runoff and lateral runoff

which are directly or in directly affected by climate

change. These changes of the climate variables were

applied to SWAT hydrological model to simulate future

water availability. The results indicate annual potential

evapotranspiration has increasing trend for both future

climate periods (Fig. 17).

Conclusions

In the present study, the physically based semi-distributed

SWAT model was used to assess the impact of climate

change on hydrological processes of Omo-Gibe river basin

in Ethiopia. The Omo-Gibe Basin was delineated into 21

sub-basins and further divided into 270 Hydrological

Response Units (HRUs) with 5% LULC and 10% soil class

cover. The sensitivity analysis has identified 7 fundamental

parameters (i.e., Cn2, ESCO, GWQMN, ALPHA_BF,

Revapmn, Ch_K2, SOL_AWC) that control the surface and

subsurface hydrological processes of the basin. However,

CN2, ESCO and ALPHA_BF were found to be highly

sensitive parameters than other parameters. The minimum

temperature varies between 0.402 and 0.137 �C and % of

Fig. 14 a Monthly percentage

change evaporation in Gibe III

and b seasonal temperature

205 Page 12 of 15 Model. Earth Syst. Environ. (2016) 2:205

123

change at 2030 s found to be 1.57 and 0.50 at 2090 s A1B

scenario. The variation of the maximum temperature varies

between 0.702 and 0.137 �C and % change at 2030s found

to be 1.57 and 0.40 �C at 2090s in Upper Omo basin. The

hydrological impact of the Omo-Gibe basin was analyzed

with respect to three ten years period baseline

(1991–2000), near term (2031–2040), end term

(2091–1000) A1B IPCC SRES climate change scenario.

The percent of mean annual change of A1B scenario

simulation on overall increasing pattern of precipitation

and evaporation from -2.58 to 2.49 at 2030 s and 5.52 to

12.35 at 2090s, respectively. The results of Omo-Gibe

(a)

(b)

Hydrological component

0200400600800

10001200140016001800

Pcp EvSR_Q

GW_QLAT_Q

PERC TLTWY

Wat

er B

alan

ce in

(mm

)

Baseline

2030s

2090s

0

5

10

15

20

25

30

Pcp Ev SR_Q GW_Q LAT_Q PERC TL TWY

Perc

enta

ge o

f Cha

nge

Fig. 15 a Water balance in the

basin and b % change in mean

annual water balance

Table 9 Percentage of change precipitation and evaporation

Date Percentage of change in precipitation Percentage of change in evaporation

2030SD% 2090SD% 2030SD% 2090SD%

Jan -22.0 -39.3 -10.98 -11.84

Feb 36.1 -1.2 21.07 6.02

Mar -12.3 -2.2 7.81 15.76

Apr -25.0 -35.7 -4.21 -0.38

May 3.7 15.5 5.33 9.93

Jun 8.0 11.6 7.89 13.80

Jul 6.0 12.7 7.77 20.35

Aug -0.8 15.2 3.66 18.72

Sep 4.4 3.2 8.05 20.73

Oct 8.2 27.2 6.46 23.87

Nov -27.4 -19.9 10.49 4.12

Dec -10.0 42.9 2.89 27.16

% mean annual change (D) -2.58 2.49 5.52 12.35

Model. Earth Syst. Environ. (2016) 2:205 Page 13 of 15 205

123

-20

-15

-10

-5

0

5

10

15

20

25

Jan

Mar

May Ju

lSe

pNo

v

Ave

Mon

thly

bas

in E

T %

of c

hang

e -40

-30

-20

-10

0

10

20

30

40

Jan

MarMay Ju

lSe

pNo

v

Ave

Mon

thly

bas

in P

CP

% o

f cha

nge

Fig. 16 Percentage change of

precipitation and evaporation

0

50

100

150

200

250

300

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Decave

mon

thly

bas

in p

reci

pita

tion

in m

m basline 2030S 2090S

0

2

4

6

8

10

12

14

16

18

20

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

ave

mon

thly

bas

in S

UR

F_Q

in m

m basline 2030S 2090S

0

5

10

15

20

25

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

ave

mon

thly

bas

in L

AT_

Q in

mm basline 2030S 2090S

0

20

40

60

80

100

120

140

160

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

ave

mon

thly

bas

in Y

IELD

_Q in

mm basline 2030S 2090S

0

10

20

30

40

50

60

70

80

90

100

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

ave

mon

thly

bas

in a

ET in

mm

basline 2030S 2090S

0

20

40

60

80

100

120

140

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

ave

mon

thly

bas

in P

ET in

mm

basline 2030S 2090S

Fig. 17 The water balance on each hydrological component

205 Page 14 of 15 Model. Earth Syst. Environ. (2016) 2:205

123

River basin study concluded that the SWAT model accu-

rately tracked the measured flows and simulated well the

monthly water yield. The study make the recommendation

that SWAT model can be effectively used for assessing the

water balance components of a river basin such as surface

runoff, lateral flow, base flow and evapotranspiration. The

simulated streamflow was found very close in agreement

with measured streamflow. Finally this study gives support

and increase awareness on the possible future impacts of

climate change on basin water balance.

Acknowledgements The authors thankfully acknowledge Min-

istry of Water and Energy, National Meteorological Agency and Min-

istry of Agriculture, Ethiopia, for providing necessary data to this study.

References

Abbaspour KC, Van Genuchten MT, Schulin R, Schlappi E (1997) A

sequential uncertainty domain inverse procedure for estimating

subsurface flow and transport parameters. Water Resour Res

33(8):1879–1892

AfB UNEP, the World Bank (2003) Poverty and climate change:

reducing the vulnerability of the poor through adaptation. In:

Sperling F (ed) African Development Bank (AfDB). United

Nations Environment Programme (UNEP) and the WB, Bonn,

p 43

Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapotranspi-

ration-guidelines for computing crop water requirements. FAO

Irrigation and Drainage Paper No. 56, Rome

Arnold JG, Fohrer N (2005) SWAT-2000 current capabilities and

research in applied watershed modeling. J Hydrol 19:53–572

Bosch DD, Sheridan JM, Batten HL, Arnold JG (2004) Evaluation of

the SWAT model on a coastal plain agricultural watershed.

Trans Am Soc Agric Eng (ASAE) 47(5):1493–1506

Chaulagain NP (2006) Impacts of climate change on water resources

of Nepal: The physical and socioeconomic dimensions. M.Sc.

thesis, University of Flensburg, Germany

Cotter AS, Chaubey I, Costello TA, Soerens TS, Nelson MA (2003)

Water quality model output uncertainty as affected by spatial

resolution of input data. J Am Water Resour Assoc

39(4):977–986

Dincer I, Midilli A, Hepbasli A, Karakoc TH (2009) Global warming:

engineering solutions, Green Energy and Technology, Springer

Science & Business Media. ISBN 1441910174, 9781441910172

Ghosh S, Mujumdar PP (2009) Climate change impact assessment:

uncertainty modeling with imprecise probability. J Geophys Res

114:D18113. doi:10.1029/2008JD011648

Intergovernmental Panel on Climate Change (IPCC) (2001) The

scientific basis: regional climate information—evaluation and

projections. http://www.ipcc.ch/ipccreports/tar/wg1/index.htm

IPCC (2001a) Climate Change (2001) Synthesis Report. In: Watson

RT, Core Writing Team (eds.) Contribution of Working Groups

I, II, and III to the Third Assessment Report of the IPCC.

Cambridge University Press, Cambridge, p 398

IPCC (2001b) Climate Change (2001): impacts, adaptation, and

vulnerability. In: McCarthy MC, Canziani OF, Leary NA,

Dokken DJ, White KS (eds) Contribution of WG II to TAR of

the Intergovernmental Panel on Climate Change. Cambridge,

p 1031

IPCC (2007) Climate Change (2007): the physical science basis. In:

Contribution of Working Group I to the Fourth Assessment

Report of the Intergovernmental Panel on Climate Change. IPCC

Kim U, Kaluarachchi JJ, Smakhtin VU (2008) Climate change

impacts on hydrology and water resources of the Upper Blue

Nile River Basin, Ethiopia. International Water Management

Institute Research Report, pp 126–127

Lenhart T, Eckhardt K, Fohrer N, Frede HG (2002) Comparison of

two different approaches of sensitivity analysis. Phys Chem

Earth 27:645–654

McSweeney C, New M, Lizcano G (2008) UNDP climate change

country profiles Ethiopia. http://country-profiles.geog.ox.ac.uk

MoWR (Ministry of Water Resources) (1996) Integrated development

of Omo-Gibe River Basin Master Plan Study, vol XI F1, F2, F3,

Addis Ababa

Nash JE, Sutcliffe JE (1970) River flow forecasting through

conceptual models—part 1-A: discussion of principles. J Hydrol

10(82):282–290

Neitsch SL, Arnold JG, Kiniry JR, Srinivasan R, Williams JR (2004)

SWAT Input/output File Documentation, Version 2005

NMA (2007) National adaptation program of action of Ethiopia

(NAPA). Final draft report. NMA, Addis Ababa

NMSA (2001) Initial National communication of Ethiopia to the

United Nations Framework convention on climate change

(UNFCCC). National Meteorological Services Agency, Addis

Ababa

Sevruk B, Matokova-Sadlonova K, Toskano L (1998) Topography

effects on small-scale precipitation variability in the Swiss pre-

Alps. Hydrology, water resources and ecology in headwaters. In:

Proceedings of the HeadWater’98 conference held at Meran/

Merano, Italy, April 1998. IAHS Publ. No. 248

Subramanya K (1998) Engineering hydrology, 2nd edn. Tata

McGraw-Hill

Temesgen TD, Hassan RM, Ringler C (2006) Measuring Ethiopian

farmers vulnerability to climate change across regional states,

Washington DC

Yang J, Reichert P, Abbaspour KC, Xia J, Yang H (2008) Comparing

uncertainty analysis techniques for a SWAT application to

Chaohe Basin in China. J Hydrol 358:1–23

Model. Earth Syst. Environ. (2016) 2:205 Page 15 of 15 205

123